Solid boundaries and mesh

[Nico Lumi]

Hi, I'm a phd student at first year and I'm start to looking at the code Basilisk for doing my simulations, but I have a few questions:

- How the boundary condition on solid boundaries are imposed ? Basically after defining a solid using the mask function the boundary condition are imposed on the nearest cell in a "classical" way, or an immersed boundary approach is used ? 

- For the mesh, I'm been able to modify the karman test case imposing some levels of refinements in the mesh (using quadtree) but when I plot the mesh using the vtk output I have only the informations (the cell values ) on the mesh initially declared.
Is it possible to obtain the values of the fields at all the levels used for the computation?

Thank you for the help!

[hud...@gmail.com]

Hi Nico,

I am also interested in solid boundaries in basilisk. It seems the no-slip boundary condition is imposed by an explicitly defined Dirichlet condition for tangent velocity of ghost cell adjacent to the boundary cell in fluid domain. I'm not sure but possibly it is one kind of (ghost-cell) immersed boundary approach. 

Best regards,

Ruifeng

[Stephane Popinet]

Hi Nico,

> - How the boundary condition on solid boundaries are imposed ? Basically
> after defining a solid using the mask function the boundary condition
> are imposed on the nearest cell in a "classical" way, or an immersed
> boundary approach is used ?

The boundary conditions on masked cells are applied in the same way as
the boundary conditions on the boundaries of the box i.e. by filling
ghost values inside the solid. The advantage is that, for cell-aligned
boundaries, the order of accuracy is the same (i.e. typically
second-order). The drawback is that this is true only for cell-aligned
boundaries.

> - For the mesh, I'm been able to modify the karman test case imposing
> some levels of refinements in the mesh (using quadtree) but when I plot
> the mesh using the vtk output I have only the informations (the cell
> values ) on the mesh initially declared.
> Is it possible to obtain the values of the fields at all the levels used
> for the computation?

Yes, but not with the VTK output.

cheers

Stephane

[h.mour]

Hello Nico,

i'm working on the same think than you .

Did you find how doing?

best regards.

[Vladimir Kolobov]

Hi Stephane,

some questions about  solids in Basilisk:

a) can you  confirm that Basilisk uses first-order stair-stepping (Lego) for
non-grid aligned solid boundaries ?

b) Can CAD files (STL/GTS) be loaded into Basilisk as in gerris ?

c) Would it be difficult to label different masks for different (solid) domains
to assign different properties and to solve (for example, heat transport) in masked domains ?

Thanks and regards,

Vladimir

[Stephane Popinet]

Hi Vladimir,

> a) can you confirm that Basilisk uses first-order stair-stepping (Lego)
> for non-grid aligned solid boundaries ?

For the moment, yes.

> b) Can CAD files (STL/GTS) be loaded into Basilisk as in gerris ?

Yes, see:

http://basilisk.fr/src/examples/distance.c

> c) Would it be difficult to label different masks for different (solid)
> domains
> to assign different properties and to solve (for example, heat
> transport) in masked domains ?

masks can be used to define different boundary conditions, not different
equations, so this would require some more coding.

cheers

Stephane

[Vladimir Kolobov]

Hi Stephane,

thanks for your response.

> a) can you  confirm that Basilisk uses first-order stair-stepping (Lego)
> for non-grid aligned solid boundaries ?

For the moment, yes.

It is our understanding that first-order stair-stepping could not give satisfactory
results for high-order moments (temperature, heat flux, etc), even with high mesh refinement levels.
Any plans for IBM implementation in the near future ?

> c) Would it be difficult to label different masks for different (solid)
> domains
> to assign different properties and to solve (for example, heat
> transport) in masked domains ?

masks can be used to define different boundary conditions, not different
equations, so this would require some more coding.

 

The ability to solve different types of equations in different parts of computational
domain would be a great advantage of Basilisk compared to Gerris.

You may want to add our recent paper http://ieeexplore.ieee.org/document/7888928/
to the gerris reference collection.

All the best,

Vladimir

[Dharmendra]

hi all


       is there any way to calculate solid surface fraction ?   

            like in case of karman test case i want to calculate surface fraction of solid cylinder at solid - fluid interface.


best

dharmendra  

[Stephane Popinet]

See:

http://basilisk.fr/src/fractions.h#computing-volume-fractions-from-a-levelset-function

[Dharmendra]

hi stephane

what i want to calculate is interface length using-

       coord m = mycs (point, f);
        double alpha = line_alpha (f[], m);
        coord p;
        length += line_length_center(m, alpha, &p);

to calculate interface length , here what i need to calculate is alpha for that .. i need  "f[]" .... i think that it will be stored somewhere as i have defined solid-fluid interface using mask function .

how can i get f[] here ?

initially what i thought was that the f[] and surface fraction in z direction will be related .

best

dharmendra

On Monday, 5 June 2017 18:53:11 UTC+5:30, Stephane Popinet wrote:

See:

http://basilisk.fr/src/fractions.h#computing-volume-fractions-from-a-levelset-function